Interesting perspective on UPST

This April 14th write up by Tiernan Ray provided some good food for thought on UPST. I don’t know, as yet, if I agree with his theses.…

I am interested in what the UPST holders on this board think.

Beachman @Iwannabeontheb2


The author states: "I’ve not seen anything convincing in the way of a third-party study that would verify the company’s mantra about how it is revolutionizing lending. So far, it really appears to be just the company’s claims.

Well OK, how exactly would a third-party actually perform such a study? Seems to me that the validation of the claim resides in Upstart’s performance as well as a number of Upstart associated banks eliminating the requirement for a prospective loan customer to hold a specific FICO score. The verification of the claim resides in the performance of Upstart. But yeah, I’ll grant, it’s not a third-party study.

Oh, BTW, do we typically see third-party studies that verify the claims of other software vendors? Are you aware of any third-party studies of Datadog’s claims? How about Zscaler? Zomminfo? I mean if third-party verification is a requirement to validate Upstart’s claims, why do all the other software companies get a pass on this? Stupid question, right? The validation resides in the performance of the software and the fact that the customers of these companies keep buying the product. But somehow, that same source of verification isn’t valid for Upstart. Give me a break.

The author goes on to say: "That’s right, half of Upstart’s opportunity is just being a new marketing channel for the banks to acquire customers. I don’t know if there’s anything AI in that, but it’s not obvious there is.

Well actually, it’s very obvious. Those referrals are a direct result of an Upstart risk analysis for that prospective customer. It’s not like, “Hey bank, here’s somebody that might be interested in becoming a customer just 'cause.” No AI analysis, no referral. It’s that simple . . . and obvious.

Finally, there’s this: “Computer models involving AI use greater precision to narrow the margin of error, thereby recommending a greater degree of risk. The thing is, all of that is formulated based on recent data. Upstart has been around since 2013, less than a decade.”

So the author simply failed to do his homework and has made a false assertion. Upstart gathered data at least as far back as 2000. Of course, their models have been under constant, iterative refinement since then. But they didn’t go into business in 2012 (despite the fact that author referenced 2013) with a totally blank slate. They had more than a solid decade of data to form the basis of their AI/ML business on day one.

The one certain thing I’ve learned from this article is that I will not be subscriber to the Technology Letter.


"I’ve not seen anything convincing in the way of a third-party study that would verify the company’s mantra about how it is revolutionizing lending. "

Good study to read on Upstart:
The graph on page 39 is interesting. I think the interest rates they are offering now are much higher now.

The author could also look at the KCAT papers that were published by KBRA and look at the CNL vs. how FICO is trending, and also compare it to there peers.

"I think it’s possible those are the questions a lot of investors are asking about the business. Of course, it’s possible I’m wrong. If I am wrong, then why is a fast-growing software company that is also profitable trading worse than many money-losing companies? "

This is a great point. Why is the market so pessimistic on Upstart right now?
Is there a fundamentally negative business development that is happening?

This is Upstart’s newest KBRA rating (as of 4/18) :…
Maybe the market is worried about the % of Grade E loans declining and the % of Grade A increasing?


There are currently a very large portion of available shares sold short. 20.8% of float is currently held short. While this has nothing to do with company performance, it is significant to price action and has been growing (up from ~18% two weeks ago).

This is OT.


Hi Brittlerrock,

Upstart gathered data at least as far back as 2000.

Can you share where you got that info? This would boost my confidence on Upstart quite a bit. Thank you.

Long Upstart


Hi Brittlerock,

Upstart gathered data at least as far back as 2000.

i would love to know where this data is obtained as well. in your previous post (…regarding) upstart, you mentioned

“Even if data from 2008 were incorporated to Upstart’s model (it may be, I don’t know if they reached that far back)”


1 Like

Thanks for the great discussion and feedback on my original question.

It is clear to me that, in this case, the author Tiernan Ray throws around a lot of conjecture, hunches and generalizations in his post. You can see him applying the same pettifogging lens in his original post from Dec 2020…when UPST when public.

I have to say that 99.99% of Tiernan’s writings are excellent. I am not unsubscribing anytime soon.

So is where I disagree with the author:

  1. We have had 5 quarterly earnings reports from UPST since it’s IPO and these have given us granular insight into the company’s performance, financials and unit metrics. If you are still persnickety about the company, then it is likely not the stock for your portfolio.

  2. An AI model that uses >1600 data points to make inferences and recommendations is nothing to sneeze at. Granted that one could feed 1600 bad data points into the model, but it would show up in terms of horrendous business results and we have not seen that with UPST. They are signing up more customers, issuing more loans and improving margins.

  3. “…as a software company, Upstart is a middle man sitting between two very large parties, loan originators, such as banks and credit unions, and parties that are willing to acquire those loans, such as the financial institutions that engage in the asset-backed securities market, where loans are bundled and securitized.”
    Ahem, this is exactly the type of fintech that I want to invest in - a technology-driven financial services company that helps its customers grow their business without taking on the debt/leverage burden on its books. UPST is way more interesting than any BPNL or insurance company…for these very reasons. 94% of their revenue comes from fees with no credit exposure. That’s just pulchritudinous!

Finally, I will throw out a couple of hunches of my own. I think that UPST is in a sweet spot right now on two fronts:

  1. Auto manufacturers have high “almost-complete” inventories on hand. Once they receive chips that are in short supply, these cars are ready to leave the factory and hit the dealer floors. New and used car prices are rolling over which will bring in new buyers, due to pent-up demand.
  2. Today mortgage rates are on the rise, however they could stabilize and come down in 2023, as interest rates normalize. UPST has plans to enter this market in 2023…could be perfect timing.

I added to my UPST position today. Stay jocose, my friends.

Beachman (@Iwannabeontheb2)


Agree with all those points and hence I fail to understand why so much pessimism for a company that strictly looking at performance from both top line and bottom line beats many of the other companies followed on this board. Guaranteed, it isnt SaaS but the valuation isnt off a SaaS business either, it did get lofty when it was in high $300s, but market cap of $7B now, I cant seem to help but wonder, what is it that we dont know or am I going to keep chasing a falling knife here (actually reminds me off opposite situation where NET was rocking to the moon without but I wasnt complaining much back then :slight_smile: and still a believer in NET)

That being said, I wanted to comment on your last point about interest rates may stabilize,…Personally I feel, that shopping for cheaper rates becomes lot.more aggressive when rates are high. When everything is low and cheap, not many people care about a mortgage at at 0.5% vs 0.51%, whereas when it’s at 3.0 vs 3.8% vs 4.2%, I would poke around and look for institutions that offer me better rate. And my guess is that, thays where UPST mortgage business helps consumers here, their algorithms finding better rates. But I do have less Hope since majority of their data is likely around sub-prime borrowing with lot higher rates? And to some degree from.the EC, David did mention that that’s the untapped market where many people with poor credit cant get mortgage just like they couldnt get personal loans and UPST solved the issue there. And so it may be a long road ahead to get any meaningful traction in this side but only time witl tell I guess. Good thing is that atleast they are improving their optionality as a business.
Anyways, coming back to my original.point. still dont why this is such a hated company :frowning: lol
Thanks all for great input.